摘要
针对目前智能交通领域中车道线检测算法效率低、鲁棒性差等问题,提出了一种基于GrowCut的车道线快速检测方法。从监控摄像机中采集图像并标定初始种子点,利用GrowCut算法进行边缘分割,对分割结果经过中值平滑滤波、边缘提取、分半处理及曲线拟合,最终得到清晰的车道线。将GrowCut算法与分水岭算法进行了对比,结果表明:该算法简便快捷、鲁棒性好,优于经典算法,可广泛应用于智能交通、公共安全领域。
A fast GrowCut-based lane detection algorithm is proposed to improve efficiency and robustness of the lane detection algorithms used in intelligent transportation.An image is captured with a surveillance camera and the initial seeds are marked on it.The GrowCut algorithm is then applied to segment the image.After median filtering,edge extraction,splitting and curve fitting,clear lanes can be obtained.The proposed algorithm is compared with the watershed method.The results show that the proposed approach has better performance and robustness than traditional algorithms.The algorithm can be applied to intelligent transportation and public security.
出处
《上海电机学院学报》
2011年第3期187-192,共6页
Journal of Shanghai Dianji University
基金
上海市教育委员会科研创新项目资助(11YZ270)
上海市高校选拔培养优秀青年教师科研专项基金项目资助(sdj10001)
关键词
车道线检测
图像分割
智能交通
lane detection
image segmentation
intelligent transportation